paraformer-speaker inference pipeline

shixian.shi

2023-10-10 8a0930d682fe3206e0b41c694fc03d7d10c7eed2

 setup.py

@@ -19,11 +19,8 @@
        "soundfile>=0.12.1",
        "h5py>=3.1.0",
        "kaldiio>=2.17.0",
        "kaldi-io==0.9.8",
        "torch_complex",
        "nltk>=3.4.5",
        "onnxruntime",
        "numexpr",
        # ASR
        "sentencepiece",
        "jieba",
@@ -37,12 +34,12 @@
        "tensorboard",
        "g2p",
        "nara_wpe",
        "Cython",
        # PAI
        "oss2",
        "edit-distance",
        "textgrid",
        "protobuf",
        "tqdm",
    ],
    # train: The modules invoked when training only.
    "train": [